Effectors and Potential Targets Selectively Upregulated in Human KRAS-Mutant Lung Adenocarcinomas

bioRxiv preprint doi: https://doi.org/10.1101/041137; this version posted June 4, 2016. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Effectors and potential targets selectively upregulated in human KRAS-mutant lung adenocarcinomas Jinyu Li1, Raffaella Sordella2, and Scott Powers1,2,3 1Department of Pathology, Stony Brook University, Stony Brook, NY, 11794 2Cold Spring Harbor Laboratory, Cold Spring Harbor, NY, 11724 3Corresponding author, email [email protected] Abstract Genetic and proteomic analysis of human tumor samples can provide an important compliment to information obtained from model systems. Here we examined protein and gene expression from the Cancer Genome and Proteome Atlases (TCGA and TCPA) to characterize proteins and protein-coding genes that are selectively upregulated in KRAS-mutant lung adenocarcinomas. Phosphoprotein activation of several MAPK signaling components was considerably stronger in KRAS-mutants than any other group of tumors, even those with activating mutations in receptor tyrosine kinases (RTKs) and BRAF. Co-occurring mutations in KRAS-mutants were associated with differential activation of PDK1 and PKC-alpha. Genes showing strong activation in RNA-seq data included negative regulators of RTK/RAF/MAPK signaling along with potential oncogenic effectors including activators of Rac and Rho proteins and the receptor protein-tyrosine phosphatase genes PTPRM and PTPRE. These results corroborate RAF/MAPK signaling as an important therapeutic target in KRAS-mutant lung adenocarcinomas and pinpoint new potential targets. Introduction How mutationally activated KRAS and other canonical RAS genes malignantly transform cells and how to block this process for therapeutic benefit has been a subject of intense investigation for over thirty years. The majority of efforts to date have relied on model systems, using established cell lines and mouse models. These studies have identified signaling pathways that are directly stimulated by biochemically active Ras proteins, including the Raf/MAPK and PI3K/Akt pathways1. They also have identified pathways or processes further downstream from Ras proteins that are involved in malignant phenotypes induced by mutant RAS genes, including the NF-κB pathway2, 3, transcriptional activity of bioRxiv preprint doi: https://doi.org/10.1101/041137; this version posted June 4, 2016. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 2 the oncogene YAP4, generation of reactive oxygen species5, and anabolic glucose metabolism6. A third class of proposed oncogenic mediators of mutant RAS genes are induced secreted proteins including TGF-α7, Vegf8, IL-89, IL-610, CXCL111, and CCL512. Several members of these three classes of mediators of oncogenic Ras have been explored as potential therapeutic targets but as of yet there hasn’t been a clinically successful treatment developed for cancers with mutant RAS genes. The recent completion of large-scale human cancer sample characterizations such as the Cancer Genome Atlas (TCGA) and the Cancer Proteome Atlas (TCPA) has enabled an altogether different approach for discovery of protein targets that are upregulated by mutant RAS genes. This approach uses a direct comparison of tumor samples containing mutant RAS genes with either corresponding normal tissue samples or with tumor samples that contain wild-type RAS genes. One chief advantage of this approach is the analysis is done using the correct in vivo physiological context, whereas with model systems there is no guarantee that the physiological context is correct although it is thought that mouse models are superior to cell culture models13, 14. Another advantage is that direct analysis of human cancer takes into account the enormous diversity of co-occurring genomic alterations. This diversity requires very large panels of human cancer cell lines to be representative and presents very difficult challenges for mouse modeling. Recently, three discrete subtypes of human KRAS-mutant lung adenocarcinomas were discovered in a breakthrough of our understanding of the variability within this type of lung cancer. These three subtypes have both distinct RNA expression profiles and different patterns of co-occurring mutations in TP53, STK11, KEAP1, and others15. Furthermore, based on experiments with appropriate cell lines, there appears to be consistent differences in responses to HSP90-inhibitors, suggesting that this classification scheme could be useful in guiding treatment strategies15. In this study, we analyzed genomic and proteomic data from human lung adenocarcinomas to examine the strength and variability of mutant-KRAS activation of proteins and genes. KRAS-mutants were compared to other lung adenocarcinoma samples and were also examined for the effects of commonly co-occurring mutations. Results Raf/MAPK signaling proteins are selectively activated in KRAS-mutant lung adenocarcinomas 230 lung adenocarcinoma samples have been comprehensively characterized for mutations and gene expression by TCGA and these same samples have been characterized by reverse-phase protein array (RPPA) for their levels of 160 different proteins and modified proteins16. To determine the selective effects of mutational activation of KRAS on these proteins in lung adenocarcinomas, we compared KRAS-mutant tumors (n= 75) to other tumors with wild-type KRAS. We further divided tumors with wild-type KRAS into those with other mutations in components of the mitogenic bioRxiv preprint doi: https://doi.org/10.1101/041137; this version posted June 4, 2016. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 3 RTK/RAF/MAPK pathway and those without, using the criteria found in the TCGA publication16. The other activating mutations in the RTK/RAF/MAPK pathway strongly tended toward mutual exclusivity and included missense or inframe deletion mutations in EGFR (n = 26), ERBB2 (n = 4), or RIT1 (n = 5); inactivating mutations in NF1 (n = 19); missense mutations in BRAF (n = 16), MAP2K1 (n = 2), HRAS (n = 1), or NRAS (n = 1); exon 14 skipping mutations in MET (n = 10), gene fusions involving ROS1 (n = 4), ALK (n = 3), or RET (n = 2); and focal high-level amplification of ERBB2 (n = 5), or MET (n = 5). We performed pairwise analysis of the three tumor subsets: KRAS mutants, other RTK/RAF/MAPK mutants, and all others. Since wanted to examine the differences with the greatest potential biological impact, we ranked the differences based on the effect size, rather than p-value which can over emphasize very small changes with little variation. We used Cohen’s d statistic that is highly related to the signal-to-noise statistic used in gene expression studies17, 18. Of all the 160 measured modified proteins and native proteins, the top ranking change in KRAS-mutant tumors when compared to either group of wild-type KRAS tumors is activation of MEK1, as judged by increased levels detected by the anti-phospho-serine 218,222 MEK1 antibody used by TCPA (Figure 1A). Also top-ranking in both comparisons is phosphorylation activation of MAPK (Figure 1B) and a direct kinase target of MAPK, p90-S6-RSK (Figure 1C). Significant large effects on activation of other direct kinase targets of MAPK (YB-1) and further downstream targets (S6) were also observed (Supplementary Table 1). Also increased in both comparisons of KRAS-mutant tumors was activation of mTOR as determined by phosphorylation at serine 2448 (Figure 1D). However, both PDK1 and AKT kinases, which might be expected to be upregulated in KRAS-mutant lung cancers due to Kras protein interaction with PI3- kinase, did not show any significant phosphorylation activation in KRAS-mutant tumors (Supplementary Table 1). Analysis of a well-validated mouse model of KRAS-mutant lung adenocarcinoma revealed that tumors showed significant activation of NF-κb and furthermore that these tumors were dependent upon NF-κb activity for tumor maintenance3. We did not observe any activation of NF-κb in human KRAS- mutant lung adenocarcinomas; in fact, NF-κb activation was significantly lower in KRAS-mutant tumors compared to other Raf/MAPK mutant tumors (Supplementary Table 1). Other mutations affecting the RTK/RAF/MAPK pathway have specific effects on protein levels We next extended our comparative protein analysis to look within the group with other mutations in the RTK/RAF/MAPK pathway. The subgroups with sufficiently large numbers for statistical analysis included EGFR mutants (n = 26), NF1 mutants (n = 19), MET mutants (n = 15), and BRAF mutants (n = 16). There were significant differences (Supplementary Table 2), the most prominent of which are highlighted in Figure 2. Notably, the phosphorylation activation of EGFR and HER2 proteins was significantly higher in EGFR mutants, but there was also increased phosphorylation of RET protein and decreased phosphorylation of HER3 and MET proteins (Figure 2). MET mutants on the other hand bioRxiv preprint doi: https://doi.org/10.1101/041137; this version posted June 4, 2016. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 4 showed opposite effects, with significant increased phosphorylation of MET protein

Load more